Proceedings of the Annual Meeting of the Cognitive Science Society

When we explain a fact or event, we typically contrast it with a specific set of counterfactual alternatives. For example, anexplanation of why Alex (as opposed to somebody else) ate the cake will seek to identify relevant factors that vary acrossagents, rather than across food items. The contrastive nature of explanation has been widely appreciated, but its cognitiveconsequence have not. We report a study with 340 adults examining how commitment to a particular explanatory contrast(agent- or patient-based) affects discovery of noisy patterns. Maximum predictive accuracy could be achieved by detectingpredictive regularities along multiple dimensions. We found that engaging in contrastive explanation (committing to aparticular contrast) impeded the discovery of alternative patterns that predicted the outcome. While explaining is likely tobe beneficial in many contexts, seeking an explanation with a single contrast could interfere with peoples ability to identifyreal structure in the world.